The chattersphere has been
busily lauding the benefits of using generative AI
tools to save time and effort by producing
ceremony scripts, blog posts, and marketing
materials.
ChatGPT and other tools draw on what are referred
to as LLMs (Large Language Models) which just
means that the algorithm processes whatever it can
get access to, effectively plagiarising largely
copyright content owned by you, me, and the world
and his wife. Ethical questions aside, is there a
downside to that?
The downside to AI
tools for Celebrants
If you look through the lens of the Code of
Practice that all Australian Authorised Marriage
Celebrants are required to comply with, yes there
is. We are required to abide by all laws, and that
means not breaching the intellectual property of
others. To be realistic, celebrants breach
copyright almost routinely through the simple act
of including passages written by others in our
ceremonies without accurate attribution. How many
times have you recognised vows, presented to you
as self-written by someone you are marrying,
because you’ve seen them time and again? Or seen a
reading attributed to Unknown or Anonymous when
the author is neither?
But neither of those breaches will grossly affect
your right to whack a © symbol on your ceremony or
blog post and assert your moral right to be
acknowledged as the author.
Nor will contracting with a human writer to
produce all or part of your content for you, as
long as your contract specifies that copyright is
transferred to you. You paid for it, so that is
legal.
Enter AI and it all changes
Where in the world you are at the time makes all
the difference because of a sneaky detail in
Australian copyright law. Who created it.
And how original is it. What are referred to
as subsistence criteria. Copyright in a work (in
this case, say, a wedding ceremony script),
subsists in the creator of the work who must have
used independent intellectual effort to create it.
Only a human being can therefore create a literary
work that is covered by Australian Copyright Law.
Game, set, match to human intelligence.
But, you might argue that you used independent
intellectual effort to create the prompt that the
AI tool used to produce the text. Good argument,
but irrelevant because any prompt is separate from
the text produced by AI. Much the same way the
exam question is separate from the student’s
answer on which the student is graded.
So, if AI produced the text, and AI isn’t human,
copyright laws don’t apply. Or do they? That’s
where it gets even more complicated.
Breaching copyright
Under Australian (and International)
copyright law, using literary works created by
others without express permission of the author,
breaches the author’s copyright, an economic right
that protects the creators right to benefit
economically from their own independent
intellectual effort. Using literary works created
by others without acknowledging them as the
creator breaches their moral rights. Copyright can
be transferred. Moral rights cannot.
A text produced by an AI tool may or may not
be in breach of copyright. If what is produced is
a synthesis of mass data, likely not. If, however,
it serves up a text that includes a large
proportion of copyright-protected material, yes it
will be
.
A text produced by an AI tool may or may not
be in breach of copyright. If what is produced is
a synthesis of mass data, likely not. If, however,
it serves up a text that includes a large
proportion of copyright-protected material, yes it
will be.
Other AI Risks
Apart from not having any
rights where the text you are using is AI
generated, are there any other risks that we
celebrants should be aware of? I suggest there are
four big ones
- AI tools can be used by anyone. So we have
no edge or advantage over a rank amateur when
it comes to AI generated ceremony creation,
which could lead to an even more widespread
belief that anyone can create and lead a
ceremony.
- LLMs do not fact check. If a false fact is
repeated often enough on machine accessible
sources (i.e. the internet) it will be served
up as fact.
- Apart from removing ourselves from any
online presence, which is neither practical
nor desirable, there is no way of protecting
our intellectual property
- There is no practical way to know, without
extensive checking of every piece of AI
generated material we might use, whether or
not it breaches copyright.
Plus there is a fifth risk, so huge that it
deserves a category all of its own. Google regards
AI-written content to be “automatically generated
content” which will cause it to flag your site.
How it does that is through the use of AI content
detectors, tools that identify both AI written
content and content written in collaboration with
human authors. Use of such tools is not restricted
to the big end of town. Within weeks of AI
generation tools bursting onto the scene late last
year, an AI Text Classifier became available, free
for all to use. Copy, paste, hit enter, and in a
nano-second be told where on a Likert scale of
Likely to Very Unlikely to have been generated
using AI, a piece of text falls. I envisage that
it will become common for clients to check for AI
content anything a celebrant claims to be an
original work purpose-written for them.
What's the solution?
Do I have a solution?
Sadly, no. AI is unregulated, fast developing, and
here to stay. There have been no controlled
“clinical” trials. So how it will all pan out is,
at the moment, anyone’s guess. In the meantime,
all we can do is to make sure, as far as we can,
that we are not in breach of the Code of Practice,
or Copyright Law as it stands.
Footnote
AI was not used in any
way in the writing of this article. But would the
classifier recognise that this article was written
by a human? I ran it through the classifier and
the result was
The classifier considers the
text to be very unlikely AI-generated.